Dating a tool

The date code people actually search for is Snap-on's, and Snap-on publish nothing about it. Not a chart, not a page, not a sentence. What exists is a collector's work, and it is very good, and it is not ours to copy.

Milwaukee, whom nobody searches for, publish theirs in full — two service bulletins, free, on a host with no restrictions at all. So this page is mostly about Milwaukee, which is not what anybody came for, and is the honest answer.

Milwaukee's serial number, field by field

Product Support Bulletin #404, August 2003, sets out how a Milwaukee serial number is built: thirteen characters, with a year and a week inside them. It is published, it is free, and it is two pages long.

Which characters, and their wording, are not reproduced here. Milwaukee's terms restrict commercial use of their content and this site is going to carry advertising, so the named fields and the plant codes have been withdrawn and permission to describe them is being asked for. The bulletin has all of it.

The week is where it stops being a date

Milwaukee give you a week, not a day — and they never say which week numbering they mean. There are two in common use and they do not agree:

Their own worked example is week 33 of 2001, which they gloss as “2^nd week of August”. Both conventions put that in mid-August, so their example does not settle which one they use.

And the gap is not small. Across the years this format has been in use, 1992 to 2026, the two conventions never once agree: in 25 of those 35 years they are one day apart, and in the other ten they are eight days apart — more than a whole week. A 1993 or 1999 tool is the bad case. The decoder above shows both ranges for the year you type and says how far apart they are; it does not average them, because the average of two conventions is not a date anybody used.

Their battery packs use seven formats, and two of them collide

Bulletin #527 is the interesting one. Milwaukee list seven different date-code layouts across their battery packs, labelled b to h. Some put the week first. Some put the year first.

So a bare four-digit code has two readings and the number cannot tell you which. Under format b the first pair is the week and the second the year. Under format g it is the other way round. The same four digits are week 7 of 2012 under one and year 2007, week 12 under the other — a five-year gap, from the same four characters, and nothing in the code says which layout the pack in your hand uses. Only the battery does.

This page does not attempt to guess between them, and neither should anything else.

Milwaukee flag one ambiguity themselves, which is rare enough on this site to be worth pointing at. One of the seven formats uses a single digit for the year, and their own key annotates it “8 = YEAR, i.e. '88, '98”. They are telling you, in the key itself, that the key does not resolve it.

Snap-on, and why this page cannot help you

Snap-on is the most-searched tool date code by a distance, and we could find nothing published by Snap-on about it. Their site carries product registration, diagnostic tool serial numbers, packaging standards for suppliers, hand tool certifications — and nothing that tells an owner how to read the code stamped on a wrench.

A search that finds nothing is a claim about the search, so here is exactly what was done: three separate searches restricted to snapon.com and its subdomains, plus the automated sweep described below, which reads a site's own sitemap and robots.txt and looks for date and serial pages in seven languages. If Snap-on do publish it somewhere, all of that missed it, and we would rather be told than guess.

What does exist is a collector's compilation, and the search results put it first for a reason: somebody has done years of careful work on it. It is not reproduced here. This site copies a manufacturer's own published figures, with attribution and permission, or it copies nothing — the guitar page removed 103 Gibson ranges on that principle, the piano page will not retype the Pierce Piano Atlas, and the watch page leaves the collector charts where they are.

If you want that information, go to the people who compiled it. They earned the traffic.

What was checked

The pool is 805 tool makers, listed in tool-makers.txt, built from Wikipedia's tool manufacturer and power tool manufacturer categories, with websites from Wikidata's official-website property rather than guessed at. A hand tool manufacturers category was tried and does not exist, so it contributed nothing; this page used to name it as a source, which claimed more than the sweep covered.

And read the swept figure knowing how much was there to read. Of the 401 counted as swept, 125 offered this method fewer than 25 pages to look at, and 107 of those offered nothing from a sitemap at all. A host that answers, allows us, and then has almost nothing to search still counts as swept. It is the number of doors opened, not the number of rooms searched.

Snap-on is in the swept set above, which is what makes the null result worth anything. Their robots.txt read cleanly with no blanket block, their sitemap was walked, and nothing matched. That is different from a maker who refused us and different again from one with no website: it is a site we were allowed to read, read, and found nothing in.

Of the 155 not probed: 62 did not answer at all, 33 returned 403 on robots.txt itself, 26 block us by name, 18 read fine and then refused the page, 12 went over our 45-second budget, and 4 disallow everybody.

Twenty-five block Claude by name — among them Estwing, Hazet, Metabo, Park Tool, SawStop, Delta Machinery, Lisle and Vaughan & Bushnell. That is the highest proportion of any family swept for this site so far. Our client does not call itself ClaudeBot and would slip past a literal reading of those files; it is honoured anyway, because a site naming Claude has said what it wants and the name on the request is not the point. Nothing was taken from any of the twenty-five.

Thirty-one candidates, and hardly any of them about a serial number

The sweep matches words for serial in seven languages, and on this family the other meanings ran riot. Agilent's hit was a video about serial dilution, a laboratory technique. Teledyne LeCroy's was SerialData, a signal-integrity product line. Desmi's was K-serien, Danish for the K series; F. Dick's was schneiden/serien, German for cutting ranges; Rösle's was gasgrill-serien; Fjällräven's was a Swedish expedition kollektion. And the American Precision Museum's was Bridgeport serial no. 1 — a specific historic machine tool, which is the one case where the phrase means exactly what it says and is still no use for dating yours.

The ones that were about dating a product mostly tell you where the number is, not how to read it: Porter-Cable, Sentry, Makino. Two went further and are worth naming even though they are not hand tools: Kidde on where the manufacture date is on a smoke alarm, and Truma on the production date of a gas pressure regulator.

Read the swept figure as swept, not as cleared. This method has demonstrably missed makers who do publish, on Zippo, Ludwig and Rheem. Milwaukee's own bulletins are a case in point: they live on documents.milwaukeetool.com, a separate host from the one in the maker list, and were found by hand.

Building the list

Makers go in by a positive test: a company is filed on Wikipedia under a category containing “companies”, “manufacturers”, “corporations”, “businesses” or “brands”, and anything with none of those is dropped and reported. 98 entries went that way.

Seven tool makers are missing from the pool because Wikipedia has no article under any name we could find: Britool, King Dick Tools, Kincrome, Teng Tools, Beta Utensili, Rigid and Record Power.

An earlier draft of this paragraph said ten, and it was wrong. The list-building tool reports names it checked and could not find, and it words that carefully: no article under the name tried. Three of them do have articles, under titles a letter or two away — Cornwell Quality Tools is filed as Cornwell Tools, Proto as Proto (tools), and Vaughan & Bushnell as Vaughan & Bushnell Manufacturing — and all three came into the pool through the categories anyway. The tool said something precise and this page restated it as something broader and false. There is now a check that compares the claim against the pool itself.

One category that does exist was deliberately not used: Garden tool manufacturers. A spade is a different question from a torque wrench, and the pool is already 722. That is a decision rather than a gap.